ADL: Very High RMSD as compared to the experimental complex conformations

rob yang nextgame at
Wed Jan 16 08:19:03 PST 2008

> I am trying to validate and reproduce the results of some of the
> experimentally known protein-ligand complexes by keeping the original
> receptor as such (original co-ordinates retained) and translating the ligand
> and then docking it back using Autodock3, but I have constantly been getting
> very High RMSD between the original and docked ligand conformations (ranging
> from 2 Angstroms to 12 Angstroms). I have been trying out playing with
> different parameters in the docking parameter file as well as the grid
> parameter file but have not been able to obatin conformations with reduced
> RMSD even after running over 100 runs of the genetic algorithm.

>From my experience, there are a few things to consider here:
1) crystal lattice effects, cofactors present, missing sidechains.
Usually I'd have to "clean up" the receptors first to add missing sidechains especially at the binding pocket. Sometimes I found that minimizing the complex from the pdb (even a short equilibration MD) can relieve some restrains of the system, and docking to the relaxed receptor can yield good results.
2) 2 A is a pretty good resolution in my mind (depending on the size of your ligand), especially considering the fact that ad3 only considers polar hydrogens and uses fixed charges. It wouldn't hurt to visualize the docked structures vs the real one, sometimes you'd be amazed how RMSD can be deceiving, but that's a different topic on its own. I've always seen docking as an approximation technique that gets the ligand to the right proximity in a reasonable conformation. Once in a while you get lucky and get it bang on, but most times you don't, and that's perfectly reasonable considering how fast it is.
3) complexity of your ligands (ie, # rotatable bonds, # atoms). The more complex it is, the larger the search space.

> I would be grateful if someone could tell something about the search space
> of ligand in Autodock3, that is how much space (what radius) is the ligand
> allowed to move once the gridcenter is defined in the grid parameter file.
> This could sort out my problem as I could restrict the ligand's movement
> into a particular volume (space) about the specified grid-center.
grid spacing is what you should play around with. ad3 lays a grid of 127x127x127 (max) pts that centres at your gridcentre. grid spacing tells it how fine the grid is (ie, distance b/w pts). My favorite is a 0.2 A spacing. Hope this helps.



More information about the autodock mailing list